Loyalty programs have evolved into essential tools for customer retention, yet traditional segmentation methods often fail to capture the complexity and dynamism of customer behaviors. This study leverages AI-driven clustering techniques to enhance loyalty program effectiveness in the banking sector. We identify actionable customer segments by applying Principal Component Analysis (PCA) for dimensionality reduction and comparing the performance of KMeans and Gaussian Mixture Models (GMM) on a large-scale transaction dataset. Our findings reveal that K-Means significantly outperforms GMM, achieving a Silhouette Score of 0.5500 versus 0.3750, indicating more clearly defined and well-separated clusters. The identified segments provide valuable insights for targeted marketing strategies, optimizing resource allocation, and improving personalized engagement. This research offers a robust framework for transitioning from static segmentation models to adaptive, data-driven customer relationship management approaches, ensuring higher retention and profitability in competitive banking environments.